Quantifying the Benefits of Link-Tracing Designs for Partnership Networks Studies
This article examines the Colorado Springs Human Immunodeficiency Virus (HIV) risk networks study, a study that used link-tracing design to assess the impact on observed network characteristics of recruiting cross-links.
Link-tracing design uses interpersonal networks to identify study participants. Examining the Colorado Springs study, the authors focused on cross-links: respondents who had at least two other respondents named as partners.
The authors estimated cross-links’ contribution to the observed network by creating a version of the network that removed all information the cross-links uniquely provided. They found that while cross-links did not identify unique individuals, they did identify unique patterns of relations among individuals.
The link-tracing techniques used in the Colorado Springs study helped find populations that would otherwise not have been included. The authors concluded that link-tracing is a practical design when studying a hard-to-reach population because cross-links can offset the limitations of convenience sampling; when it is possible to enhance link-tracing information with detail from a convenience sample; and when cross-links can provide epidemiologically relevant information for mapping risk networks.